JP6812563B2 - Systems and methods for planning radiation therapy treatment - Google Patents

Description

The present invention generally relates to the planning of radiotherapy treatments. More specifically, the present invention relates to systems and corresponding methods for planning radiotherapy treatments. The present invention also relates to computer programs and processor readable media.

In radiotherapy treatment, one or more therapeutic radiation beams are used to deliver a dose to the patient's treatment volume. Here, the orientation of each therapeutic radiation beam is an important parameter. The orientation of the therapeutic radiation beam is generally determined by the position of the radiation source, the direction of the beam center axis, and the rotation of the treatment head.

In CyberKnife ™ machines, orientation parameters can be controlled directly by a linear accelerator mounted on a robot arm. Here, it can be said that the orientations belonging to a single source position belong to a common node.

In a machine with a so-called C-arm supporting a linear accelerator, where the radiation source rotates around the patient at a constant distance between the source rotation axes, the beam orientation is the isocenter position and the rotation of the treatment gantry, the patient couch, and, respectively. It is defined by three angles representing the treatment head (colimeter angle). For machines mounted on the C-arm, it is also common to distinguish between coplanar treatment, where the couch angle is constant for all beams, and general non-coplanar treatment.

The treatment beam from the particle beam therapy machine can be guided to the patient using a rotating gantry similar to a linear accelerator mounted on a C-arm.

For delivery techniques where the beam orientation does not change during irradiation (eg, in 3D conformal radiotherapy (3D-CRT), static magnetic field intensity-modulated radiotherapy (IMRT), and intensity-modulated particle beam therapy (IMPT)), beam orientation. The set of is directly corresponding to the set of therapeutic beams.

The set of beam orientations can also be transformed into a continuous trajectory for rotational radiotherapy, such as intensity-modulated radiotherapy (VMAT), in which the beam sweeps over the patient during irradiation.

A discrete set of beam orientations is transformed into a continuous arc orbit, for example, by identifying the "shortest path" between the beam orientations in the sense that the time required to move the beam between the sets of orientations is minimized. be able to. 3D-CRT, IMRT, and VMAT are examples of different photon therapy techniques delivered using a linear accelerator mounted on a C-arm. IMPT can be delivered from a fixed beamline or by using a rotating gantry.

The general problem of selecting the appropriate subset of beam orientation from the larger candidate sets of beam orientation is not easy to solve.

Today, there are several different techniques for optimizing beam orientation, such as the so-called greedy method, which starts with beam zero and then iteratively adds one beam each time. Here, these calculations are very processing-intensive, as multiple optimizations, one for each unselected candidate beam, must be performed at each iteration, and the number of candidate beams is usually very large. Tends to be. In any case, the beam selected at each iteration is the candidate beam that gives the best improvement in the objective function value. Therefore, the greedy algorithm cannot easily provide a feasible treatment plan within a short calculation time. Computation time can be reduced by using a small number of optimization iterations for each subproblem, or even by using gradient information from only the first iteration. However, such simplification makes the beam selection fairly crude and therefore relatively poor quality.

The mixed integer programming method represents another alternative, in which case the beam orientation optimization problem is solved to global optimization by a structured enumeration of all possible solutions. Mixed integer programming is very computationally expensive and therefore not practical for practical problem sizes.

The probabilistic search method represents another example, in which the beams are randomly swapped and the switch is preserved if it improves the objective function value. This type of approach also tends to be expensive or at least time consuming. There are also metaheuristics such as simulated annealing and genetic algorithms that improve the current set of selected beams using a combination of probability and predefined search rules. Stochastic search methods and metaheuristics also require a long computational time to be able to sample most of the search space.

One further option is to employ a local search method that takes into account the continuous beam angle space and attempts to improve the objective function value by using a gradient calculated by sensitivity analysis or finite difference. Nevertheless, the local search method is only useful for fine tuning, as the beam orientation optimization problem has many minimies.

Finally, there are geometric methods that analyze the patient's anatomy and then calculate the quality score for the candidate beam direction without performing any treatment planning optimization. Geometric methods can be very fast. However, they have limited accuracy because they do not rely on actual treatment planning optimization.

Therefore, in light of the above problems and disadvantages, a widespread clinical practice is to manually select the beam orientation. Of course, this is far from ideal.

An object of the present invention is to alleviate the above problems and to be efficient and user-friendly with respect to the processing resources required to select a subset of beam orientation to use from a larger candidate set of beam orientations that may be available. It is to provide a strategy for therapy planning that enables a flexible interface.

According to one aspect of the invention, this object is achieved by a system for planning radiotherapy treatment in which the target volume is irradiated by a treatment beam. The system includes a processor and memory. The memory stores instructions that can be executed by the processor, which allows the system to operate to perform the following consecutive steps:
− A set of candidate beams, where each beam defines the composition of the treatment beam for the target volume; a treatment plan for radiotherapy using a subset of the candidate beams; an objective function that describes the quality of the treatment plan; must be met. A first that includes a feasible region that describes the requirements of the treatment plan that must be performed, and that the objective function and / or the feasible region limits the first complexity scale to less than or equal to the maximum of the first complexity. The step of getting a set of inputs that reflects the complexity criteria. The first complexity measure is related to the delivery time of the treatment plan.
-A step to perform an optimization step in which an updated treatment plan is calculated by optimizing the treatment plan with respect to the objective function and the feasible region.
-A step to check if the termination criteria are met. If satisfied, the set of selected beams is calculated based on the updated treatment plan, and the set of selected beams is a subset of the set of candidate beams. If not met, the updated treatment plan is set in the treatment plan, the updated first complexity criterion is calculated, the updated first complexity criterion is set in the first complexity criterion, and the procedure. Goes back to the optimization step for another iteration.

This system is advantageous because the proposed strategy of gradually removing the beam from the treatment plan leads to a computational process that is less computationally intensive for each iteration. Moreover, already, the first step requires relatively low computational power. Therefore, the entire planning process can be very efficient in terms of time and computational effort.

According to one embodiment of this aspect of the invention, the system further comprises at least one input interface and an output interface. At least one input interface is configured to receive objective function and feasible region definitions from local data stores or online resources, for example via manual user input. The output interface is configured to output a determined set of selected beams for presentation locally and / or remotely, for example on a display or storage. These interfaces are advantageous because they provide a high degree of flexibility in terms of the source data source and the destination data destination.

According to another embodiment of this aspect of the invention, the first complexity criterion represents a continuous differentiable approximation of the second complexity criterion. That is, the second complexity criterion limits the second complexity measure to within the maximum of the second complexity, and the second complexity measure is some of the set of candidate beams used in the treatment plan. Shows the beam. However, the second complexity criterion is itself usually a so-called step function, which is not differentiable anywhere. Therefore, the proposed first complexity criterion (based on the successive differentiable approximations described above) substantially facilitates the optimization calculation. In addition, processor-executable instructions calculate the updated second complexity criterion before returning the system to the optimization step, and the updated second complexity criterion is set to the second complexity criterion. Make it more operational to set to. This means that, like the first complexity criterion, the second complexity criterion is updated before continuing the optimization.

According to yet another embodiment of this aspect of the invention, the set of inputs further comprises the maximum of the last first complexity. Instructions that can be executed by the processor also have an updated first complexity such that the maximum of the first complexity is equal to the maximum of the last first complexity before returning to the last iteration of the optimization step. Make the system operational to calculate gender criteria. This allows us to know in advance the complexity of the resulting treatment plan.

According to yet another embodiment of this aspect of the invention, as above, the set of inputs includes the maximum of the last second complexity, and the instructions that can be executed by the processor are at the end of the optimization step. Make the system more operational to calculate an updated second complexity criterion such that the maximum of the second complexity is equal to the last maximum of the second complexity before returning to the iteration of. ..

According to a further embodiment of this aspect of the invention, the termination criteria are
[A] Maximum number of iterations of the optimization step,
[B] The first complexity measure is less than or equal to the maximum of the last first complexity,
[C] The second complexity measure is less than or equal to the maximum of the last second complexity, and [d] user input representing a stop command is received.
Based on one or more of. Therefore, there is a high degree of flexibility regarding the conditions under which the optimization process is terminated.

According to another embodiment of this aspect of the invention, the instructions executable by the processor further enable the system to perform an update process that includes the following steps before returning to the optimization step. .. Based on the updated treatment plan, a subset of candidate beam sets, the step of calculating the updated set of candidate beams, the step of setting the updated set of candidate beams to the set of candidate beams, feasible. A step of calculating an updated feasible region that reflects the requirements of the region and the requirement that only beams in the set of candidate beams can be used in the treatment plan, and a step of setting the updated feasible region as a feasible region. This means that the treatment plan is exclusively based on the beam that actually contributes to the irradiation of the treatment volume.

According to a further embodiment of this aspect of the invention, the processor-executable instructions are further updated based on the updated treatment plan and viable area before returning to the optimization step after the above update process. Make the system more operational to calculate the treatment plan. The feasible region now includes further updated treatment plans. Therefore, optimization can continue based on further updated treatment plans.

According to another embodiment of this aspect of the invention, the instructions executed by the processor depend on (a) the currently available treatment plan version before returning to the optimization step after the update process. Set the updated treatment plan in the treatment plan, or (b) make the system more operable to set the further updated treatment plan in the treatment plan.

According to yet another embodiment of this aspect of the invention, the first complexity measure of the further updated treatment regimen is less than or equal to the maximum of the first complexity before returning to the optimization step, and / Alternatively, an updated set of candidate beams is calculated such that the second complexity measure of the further updated treatment regimen is less than or equal to the maximum of the second complexity. This keeps the degree of complexity moderate throughout the optimization process.

According to a further embodiment of this aspect of the invention, an updated set of candidate beams is calculated such that the objective function value for the further updated treatment plan is as close as possible to the objective function value for the updated treatment plan. Will be done. Alternatively, the updated set of candidate beams is calculated by excluding at least one beam from the set of candidate beams. The beams excluded are the beams that contribute the least to the updated treatment plan. In both cases, rapid convergence can be expected.

According to another embodiment of this aspect of the invention, the objective function and / or feasible region describes the quality of the plan for radiotherapy treatment in terms of the deviation of the planned dose from the desired radiation dose. The planned dose is now calculated based on image data that defines the target volume received by the system, eg, via an input interface. This allows conventional CT scans or MRI data to be used for therapy planning.

According to another embodiment of this aspect of the invention, the first treatment regimen obtained in the set of inputs represents one of the following three alternatives:

The treatment scheme according to the first alternative involves irradiating the target volume with a subset of the beams in the set of candidate beams. The therapeutic beam for each beam in a subset of the candidate beam set is tailored to the target volume and has a uniform fluence profile. In addition, the fluence profile of the treatment beam for each beam in a subset of the candidate beam set is scaled so that the average planned dose for the target volume is equal to the planned target dose.

The treatment plan for the second alternative involves not delivering any irradiation, and the treatment plan for the third alternative uses a subset of the beams in the previously generated set of candidate beams for the treatment plan. Involved in delivering the irradiation.

According to yet another embodiment of this aspect of the invention, the set of candidate beams is the discretization of a continuous set of possible configurations of the therapeutic beam. The contiguous set of possible configurations is then based on the mechanical capabilities of the treatment machine, the patient's positioning with respect to the treatment machine, and / or the specifications of the desired configuration. Therefore, the appropriate parameters of the therapy and machine to be used can be added to the optimization process.

According to another aspect of the invention, this object is achieved by a method of planning a radiotherapy treatment in which the target volume is irradiated by a treatment beam. The method comprises the following consecutive steps:
− The step of obtaining a set of inputs in the processor, in this case the set of inputs, relates to a set of candidate beams and a radiotherapy treatment using a subset of the candidate beams, where each beam defines the composition of the treatment beam for the target volume. It includes an objective function that describes the treatment plan and the quality of the treatment plan, and a viable area that describes the requirements of the treatment plan that must be met. The objective function and / or the feasible region reflects a first complexity criterion that limits the first complexity measure to less than or equal to the maximum of the first complexity. The first complexity measure is then related to the delivery time of the treatment plan. In other words, the relatively low first complexity scale corresponds to treatment plans associated with relatively short delivery times and vice versa.
-A step to perform an optimization step in which an updated treatment plan is calculated by optimizing the treatment plan with respect to the objective function and the feasible region.
-A step to check if the termination criteria are met. If satisfied, the set of selected beams is calculated based on the updated treatment plan, and the set of selected beams is a subset of the set of candidate beams. If not met (ie, if the termination criteria are not met), the updated treatment plan is set in the treatment plan, the updated first complexity criterion is calculated, and the updated first complexity criterion is the first. Set to the complexity criterion of 1, the procedure returns to the optimization step described above.

The advantages of this method and its preferred embodiments can be seen in the above description relating to the proposed system.

According to a further aspect of the invention, the object is to the memory of at least one processor, including software adapted to implement the method proposed above when the program is run on at least one processor. Achieved by a loadable computer program.

According to another aspect of the invention, the object is recorded a program for controlling at least one processor to perform the method proposed above when the program is loaded on at least one processor. Achieved by processor-readable media.

One general advantage of the present invention over known measures is that all beam orientations that may be available on the machine are considered at least at a particular stage of the algorithm. The present invention also allows for easy control of calculation time.

Further advantages, beneficial features, and applications of the present invention will be seen in the following description and dependent claims.

Here, the present invention will be described in more detail by reference to preferred embodiments disclosed as examples, with reference to the accompanying drawings.

It is a figure which shows the outline of the therapy planning system which concerns on one Embodiment of this invention. FIG. 5 illustrates a first complexity scale representing a continuous differentiable function that approximates a step function representing a second complexity scale according to an embodiment of the present invention. FIG. 5 illustrates a first complexity scale representing a continuous differentiable function that approximates a step function representing a second complexity scale according to an embodiment of the present invention. It is a flow chart which illustrates the general method which concerns on this invention.

It is an object of the present invention to select a subset of the appropriate beam orientation to be used from a larger candidate set of beam orientations that may be available to provide a scheduled treatment of treatment volume.

To achieve this goal, standard requirements for treatment planning optimization must be met. Therefore, a set of parameters that describe treatment has been selected as the optimization variable. In addition, a viable set of these variables has been defined and objective functions have been formulated to quantify the quality of treatment plans. Examples of photon technology variables such as 3D-CRT, static magnetic field IMRT, and VMAT are energy fluence per pixel (surface element in beam plane), leaf position of multi-leaf collimator, dose rate and delivery time per segment, Or it is a beam weight. The standard variable for IMPT is the scan time per spot. The set of executable variables is defined by constraints that reflect the physical limits of the delivery method and may also include constraints on the planned dose distribution. Objective functions for treatment planning generally set a penalty for the deviation between the planned dose and the desired dose. However, the objective function may also take into account other aspects of treatment planning, such as the geometric complexity of the optimized collimator shape. Treatment planners generally have to be fully satisfied with the dose and variable requirements that are imposed as part of the objective function (the requirements that minimize the error) and the requirements that are imposed as constraints in the generation of the feasible region. Make choices about the requirements). Nevertheless, the exact selection of variables, feasible regions, and objective functions is not important to the strategies of the present invention.

It is also necessary to define a finite set of candidate beam orientations. Such sets can be generated in a fully automated manner or can be created based on input from the treatment planner. Possible parameters that can be used to define a finite set of candidate beam orientations are whether noncoplanar beam orientation should be allowed, whether variable collimator angles should be allowed, and possible gantry angles, couch angles, etc. And the degree of discretization with respect to the set of collimator angles. Naturally, the combination of angles that would cause a collision between the patient and the treatment head should be removed from the candidate set. The same applies to angles within the set of forbidden angles assigned by the treatment planner for other reasons.

FIG. 1 shows an overview of a system 100 according to an embodiment of the invention for planning a radiotherapy treatment in which a target volume is irradiated by a treatment beam. System 100 includes processor 110 and memory 140. There, the memory 140 stores instructions 145 (ie, software) that can be executed by the processor 110, whereby the system 100 performs successive steps [S ₁₁ ], [S ₁₂ ], and [S ₁₃ ]. Possibly it can behave to execute in a repetitive manner. However, before explaining steps [S ₁₁ ], [S ₁₂ ], and [S ₁₃ ], the rationale behind the proposed procedure is explained by mathematically explaining beam optimization.

Let | Β | be the number of elements in Β.

Further, let x = (x ₁ , ..., X _{| Β |} ) ^T be a vector of optimization variables consisting of subvectors x _b that define the optimization variables related to a given beam orientation b in Β.

X is a set of executable variable vectors (ie, the feasible region described above).

F is an objective function that takes x as a measure of planned quality, with lower values being preferred.

y = (y ₁ , ..., y _{| Β |} ) ^T is a function with component y _b , which is a measure of the contribution of x _b to the overall treatment, and is defined as follows:
If x _b has a value corresponding to b not used for treatment, then y _b (x _b ) = 0.
Otherwise, y _b (x _b )> 0.

t is a function that takes y (x) as a measure of the complexity of the plan, where t is the concentration of y (x) represented by the concentration (y (x)) (ie, zero of this vector). It is defined to depend on the number of non-elements). In the following, the proposed first complexity scale τ (x) and second complexity scale t (x) for quantifying the complexity of the plan will be described in detail.

An example of a possible scale y _b is the cumulative dose to the target volume delivered from beam b, the beam-on time for beam b, or a particular norm of vector x _b . The function t can be, for example, a concentration of y, or a particular norm of concentration y + y.

The first example corresponds to the case where some static beams should be controlled, the second example corresponds to the case where the total treatment time should be controlled, and the concentration term further Represents a constant time cost for a static beam, and the norm term represents the beam-on time for such a beam. A possible definition of t for rotational radiotherapy is the length of the shortest viable pathway (or the length of the shortest viable pathway connecting the subsets of beam orientation { _b ∈ Β: y _b (x _b )> 0} used for treatment. A specific approximation of the length of such a path). The definition of t for CyberKnife ™ treatment is a measure of the concentration of the number of nodes in which the first component is used for treatment and the concentration of the number of beam orientations in which the second component is used for treatment. It is a vector, which is a scale.

Beam orientation optimization can now be described as a problem:
Minimize F (x) under the condition x ∈ X, t (x) ≤ t _max (P1)
In the equation, t minimizes F (x) + G (t (x)), either within the maximum t _max of a particular acceptable complexity, or under the condition x ∈ X (P2).
In the formula, G is a penalty for t (x), eg,
With respect to the positive weight λ, G (t (x)) = λ (max {t (x) −t _max , 0}.

The solution x ^* to (P1) or (P2) defines the set of selected beam orientations Β ^* , where Β ^* = { _b ∈ Β: y _b (x ^* _b )> 0}. ..

However, the optimization problems (P1) and (P2) are very difficult to solve because the concentration of y must be controlled during the optimization. It is known that such a general convex concentration problem becomes nondeterministic polynomial time difficulty (NP-hard).

Therefore, according to the present invention, heuristics are instead used to find an approximate solution to the problem (P1) or (P2) within a practical calculation time. Heuristics approximate t by a continuous differentiable function τ _ε that is better suited for optimization. Here, the subscript ε indicates a parameter that controls the accuracy of the approximation.

Heuristics also sequentially remove beams from a set of candidate beams Β to a subset Β ^* of selected beams. More precisely, the set of candidate beams Β defines Β ^* as a subset of the selected beams in Β, and then replaces the previous set Β of the candidate beams with the subset Β ^* of the selected beams. Updated by.

The beam orientation thus removed is one that is constrained to give zero or at least inadequate contribution to treatment.

という条件の下でＦ（ｘ）を最小化する （Ａ１）
式中、

は、複雑性限界である、及び
以下の形態の（Ｐ２）の近似カウンターパート：

という条件の下でＦ（ｘ）＋Ｇ（τ_ε（ｘ））を最小化する （Ａ２）
を提供する。 Both, this is the approximate counterpart of (P1) of the form:

F (x) is minimized under the condition (A1)
During the ceremony

Is the complexity limit, and the approximate counterpart of (P2) of the form:

Under the condition that F (x) + G (τ _ε (x)) is minimized (A2)
I will provide a.

に従って濃度（ｚ）を階段関数ｓの和として表すことができることと、階段関数ｓを連続関数により近似することができることを用いることにより、ｎ個のベクトルの濃度ｚの連続近似を得ることが可能である。ここで、濃度（・）がｙに適用されるときｎ＝｜Β｜であることに留意されたい。

By using the fact that the density (z) can be expressed as the sum of the step functions s according to the above and that the step function s can be approximated by a continuous function, it is possible to obtain a continuous approximation of the density z of n vectors. Is. Note that n = | Β | when the concentration (.) Is applied to y.

2a and 2b show first and second examples of continuous approximation of smooth density (y) in the positive region.

図２ｂは、次式によりｙのεノルムを表す：

More precisely, FIG. 2a represents the sum of logistic functions by the following equation:

FIG. 2b represents the ε norm of y by the following equation:

Both of these approximations approach the concentration (y) as ε approaches zero, as can be seen in FIGS. 2a and 2b, which show the inner sum factors in (A1) and (A2), respectively.

Further, the optimization problems in (A1) and (A2) are continuous optimization problems to which many general-purpose solvers can be applied.

As will be described later, an example of a termination criterion for optimization is that the current iteration x has converged in a sense, or that some form of timeout has occurred, eg, the maximum number of iterations has been reached. Is. The initial point for optimization can in any case be a preselected initial estimate (eg, scale a uniform fluence profile so that the target dose is equal to the prescribed dose on average). ).

In light of the above, the following procedure according to the present invention applies. Specifically, instructions 145 in memory 140 that can be executed by the processor 110 enable the system 100 to perform successive steps [S ₁₁ ], [S ₁₂ ], and [S ₁₃ ]. ..

In step _{[S 11],} or via one or more interfaces 120 from internal storage system 100, the set of input can be obtained.

The set of inputs contains a set of candidate beams Β, and each beam in the set of candidate beams Β defines the composition of the therapeutic beam for the target volume. The set of inputs also includes a treatment plan x for radiotherapy treatment using a subset of candidate beams Β. In addition, the set of inputs includes an objective function F that describes the quality of the treatment plan x and a feasible region X that describes the requirements of the treatment plan x that must be met.

が予定標的線量に等しくなるようにさらにスケール変更される。 The treatment plan x obtained in the set of inputs can represent the irradiation of the target volume with a subset of the beams in the set of candidate beams Β. The therapeutic beam for each beam in the subset of the candidate beam set Β is tailored to the target volume and has a uniform fluence profile. The fluence profile of the therapeutic beam for each beam in the subset of the candidate beam set Β is the average planned dose for the target volume.

Is further scaled to be equal to the planned target dose.

Alternatively, the treatment plan x obtained in the set of inputs does not deliver the irradiation, or delivers the irradiation using a subset of the beams in the previously generated set of candidate beams Β for the treatment plan. Can be represented.

以下となるように制限する第１の複雑性基準

を反映する。第１の複雑性尺度τ（ｘ）は、治療計画ｘの送達時間に関係する。これは、比較的低い第１の複雑性尺度が、比較的短い送達時間に関連する治療計画に対応することを意味し、逆もまた同様である。 The objective function F and / or the feasible region X sets the first complexity scale τ (x) to the maximum of the first complexity.

First complexity criterion limiting to:

To reflect. The first complexity measure τ (x) is related to the delivery time of treatment plan x. This means that the relatively low first complexity scale corresponds to treatment plans associated with relatively short delivery times, and vice versa.

The objective function F and / or the feasible region X also preferably describes the quality of the plan for radiation therapy treatment in terms of the deviation of the planned dose D from the desired radiation dose D _ref . The planned dose D is then calculated based on the image data _Dimg that defines the target volume. The image data _Dimg may represent, for example, CT scan data or MRI data.

To reduce the amount of computation, the set of candidate beams Β is preferably a discretization of a continuous set of possible configurations of the therapeutic beam. A contiguous set of possible configurations may then be based on the multiple mechanical capabilities of the treatment machine, the patient's positioning with respect to the treatment machine, and / or the specifications of the desired configuration.

In step [S ₁₂ ] following step [S ₁₁ ], the optimization step in which the processor 110 calculates the updated treatment plan x'by optimizing the treatment plan x with respect to the objective function F and the feasible region X. Is executed. After that, step [S ₁₃ ] is performed.

In step [S ₁₃ ], it is checked whether the termination criteria are met, and if so, the set of beams Β ^* selected in step [S ₁₃₁ ] is calculated. The set of selected beams Β ^* is calculated based on the updated treatment plan x'. The selected beam set Β ^* is a subset of the candidate beam set Β. That is, the set of selected beams Β ^* contains less or at most the same number of beams than the set of candidate beams Β.

も計算され、更新された第１の複雑性基準

は、第１の複雑性基準

に設定される。その後、手順は、さらなる最適化のためにステップ［Ｓ_１２］に戻る。 If, in step [S ₁₃ ], it is found that the termination criteria are not met, the updated treatment plan x'is set in the treatment plan x, following step [S ₁₃₂ ]. In step [S ₁₃₂ ], the updated first complexity criterion

Also calculated and updated the first complexity criterion

Is the first complexity criterion

Is set to. The procedure then returns to step [S ₁₂ ] for further optimization.

Preferably, the system 100 further includes at least one input interface 120 and an output interface 130. The at least one input interface 120 is configured to receive the definitions of the objective function F and the feasible region X, for example, from the local data store 153 or from the online resource 152. Of course, the definitions of the objective function F and the feasible region X may be input via manual user input through a keyboard 151, a touch screen, or similar input members. In addition, at least one input interface 120 provides a description of the desired radiation dose _Dref , eg, via the keyboard 151 or the online resource 152, and / or a target volume, eg, from the online resource 152 or the local data store 153. The image data _Dimg to be defined may be received. The output interface 130 is configured to output a determined set of selected beams Β ^* for storage in the local data store 163 and / or the remote data resource 162, for example, for presentation on display 161. ..

は、好ましくは、第２の複雑性基準

の連続する微分可能な近似を表す。すなわち、このような連続する微分可能な近似は、基本的なＮＰ困難問題についてなされる対応する計算に実質的に関係する最適化に関連した計算を簡易化する。第１の複雑性基準

と同様に、

は、第２の複雑性尺度ｔ（ｘ）を第２の複雑性の最大

以内に制限する。第２の複雑性尺度ｔ（ｘ）は、ここで、治療計画ｘに用いられる候補ビームの集合Βにおけるいくつかのビームを示す。 As mentioned above, the first complexity criterion

Is preferably a second complexity criterion

Represents a continuous differentiable approximation of. That is, such a continuous differentiable approximation simplifies the optimization-related calculations that are substantially relevant to the corresponding calculations made on the basic NP-hard problem. First complexity criterion

alike,

Sets the second complexity measure t (x) to the maximum of the second complexity

Limit within. The second complexity measure t (x) shows here some beams in the set of candidate beams Β used in the treatment plan x.

を計算し、更新された第２の複雑性基準

を第２の複雑性基準

に設定するようにシステム１００をさらに動作可能にする。 To gradually increase the treatment plan, instructions 145 executable by processor 110, before returning to step [S _12], a second complexity criteria are updated

Calculated and updated second complexity criterion

The second complexity criterion

Make the system 100 more operable to set to.

が最後の第１の複雑性の最大τ_ｍａｘに等しくなるような更新された第１の複雑性基準

を計算するようにシステム１００をさらに動作可能にする。 To ensure reasonable convergence of the optimization, the set of inputs preferably further includes a maximum of τ _max of the last first complexity, and the instructions 145 executable by processor 110 are in step [S ₁₂ ]. Maximum of the first complexity before returning to the last iteration of

Updated first complexity criterion such that is equal to the maximum τ _max of the last first complexity

Make the system 100 more operable to calculate.

が最後の第２の複雑性の最大ｔ_ｍａｘに等しくなるような更新された第２の複雑性基準

を計算するようにシステム１００を動作可能にする。 Alternatively, the set of inputs contains the maximum τ _max of the last first complexity, but the instruction 145 executable by processor 110 instead returns to the last iteration of step [S ₁₂ ]. Maximum of second complexity

Updated second complexity criterion such that is equal to the maximum t _max of the last second complexity

Make the system 100 operational so as to calculate.

According to one embodiment of the invention, the termination criterion is the maximum number of iterations of step [S ₁₂ ], where the first complexity scale τ (x) is less than or equal to the last first complexity maximum τ _max. It is based on the fact that the second complexity measure t (x) is less than or equal to the maximum t _max of the last second complexity, and / or the user input S representing the stop command is received. Thereby, the optimization process can be terminated when certain criteria are met and at any time selected by the operator, eg, the treatment planner.

Preferably, instructions 145 executable by processor 110, before returning to step _{[S 12],} further operable to system 100 to perform the update process. The update process then involves the following steps:

The step of calculating the updated set Β'of candidate beams based on the updated treatment plan x'. The updated set Β'of the candidate beams is here a subset of the set Β of the candidate beams.

The updated set Β'of the candidate beams is then set to the set Β of the candidate beams, the updated feasible region X'is calculated, and the updated feasible region X'is the requirements of the feasible region X and It reflects the requirement that only the beams in the set of candidate beams Β can be used in treatment plan x. The updated feasible region X'is then set in the feasible region X, and the procedure can loop back to step [S ₁₂ ] for further optimization.

Nevertheless, before returning to step [S _12] After the update process, the processor 110 also preferably updated treatment plan x 'and the feasible region X to a further updated treatment plan x based''Calculate. Here, the feasible region X includes a further updated treatment plan x''.

More preferably, the candidate beam is updated so that the objective function value F (x'') for the further updated treatment plan x'' is as close as possible to the objective function value F (x') for the updated treatment plan x'. The set Β'is calculated. That is, it accelerates the convergence of optimizations.

In addition, calculating the updated set Β'of candidate beams involves excluding at least one beam from the set Β of candidate beams, and at least one beam so excluded is updated. It is advantageous to have at least one beam that contributes the least to the treatment plan x'.

Further, before returning to step [S ₁₂ ] after the above update process, the processor 110 preferably sets the updated treatment plan x'to the treatment plan x, or further updated treatment plan x. Perform the step of setting'' to treatment plan x. This improves the quality of the treatment plan obtained from the last iteration of the optimization step.

以下である、及び
［ｉｉ］さらなる更新された治療計画ｘ’’の第２の複雑性尺度ｔ（ｘ’’）が第２の複雑性の最大

以下である。 According to one embodiment of the invention, before returning to step [S ₁₂ ], an updated set Β'of candidate beams is calculated such that one or both of the following conditions are satisfied.
[I] The first complexity measure τ (x'') of the further updated treatment plan x'' is the maximum of the first complexity.

The second complexity measure t (x'') of the following, and [ii] further updated treatment plan x'' is the maximum of the second complexity.

It is as follows.

To summarize, with reference to the flow chart of FIG. 3, a general method according to the present invention for planning a radiotherapy treatment in which the target volume is irradiated by the treatment beam will be described here.

以下となるように制限する第１の複雑性基準

を反映する。第１の複雑性尺度τ（ｘ）は、治療計画ｘの送達時間に関係する。 In a first step _{S 11,} the set of input in the processor 110 can be obtained. The set of inputs contains a set of candidate beams Β, and each beam defines the composition of the therapeutic beam for the target volume. The set of inputs also includes a treatment plan x for radiotherapy treatment, which uses a subset of the candidate beam Β. In addition, the set of inputs includes an objective function F that describes the quality of the treatment plan x and a feasible region X that describes the requirements of the treatment plan x that must be met. The objective function F and / or the feasible region X sets the first complexity scale τ (x) to the maximum of the first complexity.

First complexity criterion limiting to:

To reflect. The first complexity measure τ (x) is related to the delivery time of treatment plan x.

Then, in step S _12, the updated treatment plan x 'is calculated by optimizing the treatment plan x respect the objective function F and feasible region X, the optimization step is performed.

も計算され、更新された第１の複雑性基準

が第１の複雑性基準

に設定される。その後、手順は、さらなる最適化のためにステップＳ_１２にループバックする。 Thereafter, in step S _13, to check whether a termination criterion is met, when satisfied, step S ₁₃₁ is followed. If not met, the procedure continues to step S _132, where the updated treatment plan x 'is set in the treatment plan x. In step _S131 , the updated first complexity criterion

Also calculated and updated the first complexity criterion

Is the first complexity criterion

Is set to. Thereafter, the procedure loops back to step S ₁₂ for further optimization.

In step S _131, a set of beams selected based on the updated treatment plan x 'beta ^* is calculated. The selected beam set Β ^* is a subset of the candidate beam set Β. The procedure is then completed and radiotherapy treatment can be presented to the treatment planner, whose treatment is based on the selected beam set Β ^* and the updated treatment plan x'.

All of the process steps described with reference to FIG. 3 and any subsequence of steps can be controlled by a programmed processor. Further, although the embodiments of the present invention described above with reference to the drawings include a processor and a process performed by at least one processor, the present invention is adapted to implement computer programs, in particular the present invention. It also extends to computer programs on or within the carrier. The program can be an intermediate form between the source code and the object code, such as source code, object code, a partially compiled form, or any other form suitable for use in implementing the processes according to the invention. .. The program may be part of the operating system or may be a separate application. The carrier can be any entity or device that can carry the program. For example, the carrier may be a flash memory, a ROM (Read Only Memory), for example, a DVD (Digital Video / Versaille Disk), a CD (Compact Disk), or a semiconductor ROM, an EPROM (Erasable Programmable Read-Ollemagry Memory). It may include a storage medium such as a Read-Only Memory) or a magnetic recording medium such as a floppy disk or a hard disk. Further, the carrier can be a transmittable carrier such as an electrical or optical signal that can be transmitted via an electrical or optical cable or by radio or other means. When the program is embodied in a signal that can be transmitted directly by a cable or other device or means, the carrier may be composed of such a cable or device or means. Alternatively, the carrier may be an integrated circuit into which the program is incorporated, and the integrated circuit is adapted for use in performing or performing the relevant process.

As used herein, the term "preparing, including" is adopted to specify the presence of a described feature, integer, step, or component. However, the term does not exclude the presence or addition of one or more additional features, integers, steps, or components, or groups thereof.

The present invention is not limited to the embodiments described in the drawings and may be freely modified within the scope of the claims.

Claims (30)

A system (100) for planning radiotherapy treatment in which the target volume is irradiated by a treatment beam.
With the processor (110)
A memory (140) for storing an instruction (145) that can be executed by the processor (110), and
The system (100), thereby
[S ₁₁ ] A step of acquiring a set of inputs.
The set of inputs is
A set of candidate beams (Β), each of which defines the composition of the therapeutic beam with respect to the target volume.
A treatment plan (x) for the radiotherapy treatment using the subset of the candidate beam (Β), and
The objective function (F) that describes the quality of the treatment plan (x) and
A feasible region (X) that describes the requirements of the treatment plan (x) that must be met, and
And at least one of the objective function (F) and the feasible region (X) is the first complexity criterion.

Reflecting the above, the first complexity criterion

The first complexity measure (τ (x)) is the step of obtaining a set of inputs related to the delivery time with respect to the treatment plan (x), limited to:
[S ₁₂ ] Performing an optimization step in which an updated treatment plan (x') is calculated by optimizing the treatment plan (x) with respect to the objective function (F) and the feasible region (X). Steps to do and
[S ₁₃ ] Check if the termination criteria are met and
If satisfied, [S ₁₃₁ ] a subset of the candidate beam set (Β), the selected beam set (Β ^* ), is calculated based on the updated treatment plan (x').
If not met, [S ₁₃₂ ] the updated treatment plan (x') is set to the treatment plan (x) and the updated first complexity criterion is set.

And the updated first complexity criterion

The first complexity criterion

Set to and return to step [S ₁₂ ], and
A system (100) that is capable of operating to perform successive steps, including
The first complexity criterion

But the second complexity criterion

Represents a continuous differentiable approximation of, said second complexity criterion

Sets the second complexity scale (t (x)) to the maximum of the second complexity

Limited to within, the second complexity measure (t (x)) indicates some beams in the set of candidate beams (Β) used in said treatment plan (x), said processor (110). The second complexity criterion updated by the executable instruction (145) before returning to step [S ₁₂ ].

And the updated second complexity criterion

The second complexity criterion

The system (100), which makes the system (100) more operable to be set to . With at least one input interface (120) configured to receive the definitions of the objective function (F) and the feasible region (X).
An output interface (130) configured to output the determined set of selected beams (Β ^* ), and
The system (100) according to claim 1, further comprising. The set of inputs further includes the final second complexity _max (τ _max ), before the instruction (145) executable by the processor (110) returns to the last iteration of step [S ₁₂ ]. In addition, the maximum of the first complexity

Is an updated first complexity criterion such that is equal to the last first complexity maximum (τ _max ).

The system (100) according to claim 1 or 2 , further making the system (100) operational so as to calculate. The set of inputs further comprises a final second complexity _max (t _max ), and the processor (110) returns the executable instruction (145) to the last iteration of step [S ₁₂ ]. Before, the maximum of the second complexity

Is an updated second complexity criterion such that is equal to the last second complexity maximum (t _max ).

The system (100) according to claim 1, further making the system (100) operational so as to calculate. The termination criteria
[A] Maximum number of iterations of step [S ₁₂ ],
[B] The first complexity scale (τ (x)) is less than or equal to the maximum (τ _max ) of the last first complexity.
[C] The second complexity scale (t (x)) is less than or equal to the maximum (t _max ) of the last second complexity, and [d] user input (S) representing a stop command is received. Be done,
The system (100) of claim 4 , which is based on at least one of. Before the instruction (145) that can be executed by the processor (110) returns to step [S ₁₂ ],
Based on the updated treatment plan (x'), an updated set of candidate beams (Β'), which is a subset of the set of candidate beams (Β), is calculated.
The updated set of candidate beams (Β') is set as the set of candidate beams (Β),
An updated feasible region (X') that reflects the requirement of the feasible region (X) and the requirement that only the beams in the set of candidate beams (Β) can be used in the treatment plan (x) is calculated.
Setting the updated feasible region (X') as the feasible region (X),
The system (100) according to any one of claims 1 to 5 , further enabling the system (100) to perform an update process including. Executable said instructions by said processor (110) (145), before the after the update process returns to step [S _12], the updated treatment plan (x ') and the execution area (X) The system (100) is further operational to calculate a further updated treatment plan (x'') based on the executable area (X) with the further updated treatment plan (x''). The system (100) according to claim 6, which includes. Executable said instructions by said processor (110) (145), before returning to step _{[S 12]} After the update process,
A step of setting the updated treatment plan (x') in the treatment plan (x), or a step of setting the further updated treatment plan (x'') in the treatment plan (x).
The system (100) according to any one of claims 6 or 7 , further enabling the system (100) to operate one of the two. Before returning to step [S ₁₂ ]
[I] The first complexity measure (τ (x'')) of the further updated treatment plan (x'') is the maximum of the first complexity.

And [ii] the second complexity measure (t (x'')) of the further updated treatment plan (x'') is the maximum of the second complexity.

Is below
The system (100) according to any one of claims 6 to 8 , wherein an updated set (Β') of the candidate beams is calculated so that at least one of them is satisfied. The objective function value (F (x'')) for the further updated treatment plan (x'') is as close as possible to the objective function value (F (x')) for the updated treatment plan (x'). The system (100) according to claim 7 , wherein an updated set (Β') of such candidate beams is calculated. By excluding at least one beam from the set of candidate beams (Β) that the instruction (145) that can be executed by the processor (110) contributes the least to the updated treatment plan (x'). The system (100) according to any one of claims 6 to 10 , further enabling the system (100) to calculate an updated set (Β') of the candidate beams. At least one of the objective function (F) and the viable region (X) determines the quality of the plan for radiotherapy treatment in terms of the deviation of the planned dose (D) from the desired radiation dose ( _Dref ). The system (100) according to any one of claims 1 to 11 , wherein the planned dose (D) is calculated based on the image data ( _Dimg ) defining the target volume. The treatment plan (x) obtained in the set of inputs is
Irradiating the target volume with a subset of the beams in the candidate beam set (Β).
The treatment beam for each beam in the subset of the candidate beam set (Β) is tailored to the target volume, has a uniform fluence profile, and relates to each beam in the subset of the candidate beam set (Β). The fluence profile of the therapeutic beam is the average planned dose for said target volume.

Irradiating the target volume with a subset of the beams in the candidate beam set (Β), which is scaled to be equal to the planned target dose.
Not delivering irradiation, or delivering irradiation using a subset of the beams in the previously generated set of candidate beams (Β) for the treatment plan.
The system (100) according to any one of claims 1 to 12 , which represents one of the above. The set of candidate beams (Β) is a discretization of a continuous set of possible configurations of the therapeutic beam, and the continuous set of possible configurations is
Multiple mechanical capabilities of the treatment machine,
Positioning of the patient with respect to the treatment machine, and specifications of the desired configuration,
The system (100) according to any one of claims 1 to 13 , which is based on at least one of the two. A method of planning a radiotherapy treatment in which a target volume is irradiated by a treatment beam, comprising a series of steps, wherein the succession steps.
[S ₁₁ ] A step of acquiring a set of inputs in the processor (110).
The set of inputs is
A set of candidate beams (Β), each of which defines the composition of the therapeutic beam with respect to the target volume.
A treatment plan (x) for the radiotherapy treatment using the subset of the candidate beam (Β), and
The objective function (F) that describes the quality of the treatment plan (x) and
A feasible region (X) that describes the requirements of the treatment plan (x) that must be met,
And at least one of the objective function (F) and the feasible region (X) is the first complexity criterion.

Reflecting the above, the first complexity criterion

The first complexity scale (τ (x)) is the maximum of the first complexity

The first complexity measure (τ (x)) is the step of obtaining a set of inputs related to the delivery time with respect to the treatment plan (x), limited to:
[S ₁₂ ] Performing an optimization step in which an updated treatment plan (x') is calculated by optimizing the treatment plan (x) with respect to the objective function (F) and the feasible region (X). Steps to do and
[S ₁₃ ] Check if the termination criteria are met and
If satisfied, [S ₁₃₁ ] a subset of the candidate beam set (Β), the selected beam set (Β ^* ), is calculated based on the updated treatment plan (x').
If not met, [S ₁₃₂ ] the updated treatment plan (x') is set to the treatment plan (x) and the updated first complexity criterion is set.

And the updated first complexity criterion

The first complexity criterion

Set to and return to step [S ₁₂ ], and
Is a method , including
The first complexity criterion

But the second complexity criterion

Represents a continuous differentiable approximation of, said second complexity criterion

Sets the second complexity scale (t (x)) to the maximum of the second complexity

Limited to within, the second complexity measure (t (x)) indicates some beams in the set of candidate beams (Β) used in said treatment plan (x), the method of stepping. Updated second complexity criterion before returning to [S ₁₂ ]

Calculate and
The updated second complexity criterion

The second complexity criterion

To set to
A method that further includes . A step of receiving the definitions of the objective function (F) and the feasible region (X) via at least one input interface (120).
A step of outputting the determined set of selected beams (Β ^* ) via the output interface (130), and
15. The method of claim 15 . The set of inputs further comprises the final first complexity _max (τ _max ), and the first complexity _max before the method returns to the last iteration of step [S ₁₂ ].

Is equal to the maximum of the last first complexity (τ _max ) of the updated first complexity criterion.

To calculate,
15. The method of claim 15 or 16 , further comprising. The set of inputs further comprises a final second complexity _max (t _max ), and the second complexity _max before the method returns to the last iteration of step [S ₁₂ ].

Is equal to the maximum (t _max ) of the last second complexity of the updated second complexity criterion.

15. The method of claim 15 , further comprising calculating. The termination criteria
[A] Maximum number of iterations of step [S ₁₂ ],
[B] The first complexity scale (τ (x)) is less than or equal to the maximum (τ _max ) of the last first complexity.
[C] The second complexity scale (t (x)) is less than or equal to the maximum (t _max ) of the last second complexity, and [d] user input (S) representing a stop command is received. Be done,
18. The method of claim 18 , which is based on at least one of. Before returning to step [S ₁₂ ]
Based on the updated treatment plan (x'), an updated set of candidate beams (Β'), which is a subset of the set of candidate beams (Β), is calculated.
The updated set of candidate beams (Β') is set as the set of candidate beams (Β),
Calculate the updated feasible region (X') that reflects the requirement of the feasible region (X) and the requirement that only the beams in the set of candidate beams (Β) can be used in the treatment plan (x). Setting the updated feasible region (X') to the feasible region (X),
Further comprising a method according to any one of claims 15 to claim 19 to perform the update process that involves. A further updated treatment plan (x'') is calculated based on the updated treatment plan (x') and the feasible region (X) after the update process and before returning to step [S ₁₂ ]. 20. The method of claim 20 , wherein the feasible region (X) comprises the further updated treatment plan (x ″). Before after the update process returns to the step _{[S 12],}
A step of setting the updated treatment plan (x') in the treatment plan (x), or a step of setting the further updated treatment plan (x'') in the treatment plan (x).
The method of any one of claims 20 or 21 , further comprising performing one of the following. Before returning to step [S ₁₂ ]
[I] The first complexity measure (τ (x'')) of the further updated treatment plan (x'') is the maximum of the first complexity.

And [ii] the second complexity measure (t (x'')) of the further updated treatment plan (x'') is the maximum of the second complexity.

Is below
The method of any one of claims 20 to 22 , comprising calculating an updated set (Β') of said candidate beams such that at least one of them is satisfied. The objective function value (F (x'')) for the further updated treatment plan (x'') is as close as possible to the objective function value (F (x')) for the updated treatment plan (x'). 23. The method of claim 23 , comprising calculating an updated set (Β') of said candidate beams such as. Calculating the updated set of candidate beams (Β') by excluding at least one beam that contributes the least to the updated treatment plan (x') from the set of candidate beams (Β). The method according to any one of claims 19 to 24 , further comprising. At least one of the objective function (F) and the feasible region (X) describes the quality of the radiation therapy treatment plan in terms of the deviation of the planned dose (D) from the desired radiation dose ( _Dref ). The method according to any one of claims 15 to 25 , wherein the method comprises calculating a planned dose (D) based on image data ( _Dimg ) defining a target volume. The treatment plan (x) obtained in the set of inputs is
Irradiating the target volume using a subset of the beams in the candidate beam set (Β).
The treatment beam for each beam in the subset of the candidate beam set (Β) is tailored to the target volume and has a uniform fluence profile, and the treatment for each beam in the subset of the candidate beam set (Β). The beam fluence profile is the average planned dose for the target volume.

Irradiating the target volume, which is scaled to be equal to the planned target dose,
Not delivering irradiation, or delivering irradiation using a subset of the beams in the previously generated set of candidate beams (Β) for the treatment plan.
The method according to any one of claims 15 to 26 , which represents one of the above. The set of candidate beams (Β) is a discretization of a continuous set of possible configurations of the therapeutic beam, and the continuous set of possible configurations is
Multiple mechanical capabilities of the treatment machine,
Positioning of the patient with respect to the treatment machine, and specifications of the desired configuration,
The method according to any one of claims 15 to 27 , which is based on at least one of the two. A computer program (145) comprising an instruction to cause the at least one processor (110) to execute the method according to any one of claims 15 to 28 when executed on at least one processor (110). ). A non-volatile data carrier (140) that stores the computer program (145) of claim 29 .

Images